Download Eco-directed sustainable prescribing

Science of the Total Environment 493 (2014) 392–404
Contents lists available at ScienceDirect
Science of the Total Environment
journal homepage: www.elsevier.com/locate/scitotenv
Eco-directed sustainable prescribing: feasibility for reducing water
contamination by drugs
Christian G. Daughton ⁎
Environmental Sciences Division, National Exposure Research Laboratory, U.S. Environmental Protection Agency, 944 East Harmon Ave, Las Vegas, NV 89119, United States
H I G H L I G H T S
•
•
•
•
•
Role of pollution prevention is examined for reducing drug entry to the environment.
Eco-directed sustainable prescribing (EDSP) is proposed for reducing drug excretion.
Drug loadings in environment via sewers are dictated by pharmacokinetics.
Prescribing could be guided by selecting drugs that are poorly excreted.
Do empirical environmental occurrence data for drugs correlate with pharmacokinetics?
a r t i c l e
i n f o
Article history:
Received 1 May 2014
Received in revised form 3 June 2014
Accepted 3 June 2014
Available online xxxx
Editor: Damia Barcelo
Keywords:
Drugs
Active pharmaceutical ingredients
Pharmacokinetics
Prescribing
Water pollution
Sustainable healthcare
a b s t r a c t
Active pharmaceutical ingredients (APIs) from the purchase and use of medications are recognized as ubiquitous
contaminants of the environment. Ecological impacts can range from subtle to overt — resulting from multigenerational chronic exposure to trace levels of multiple APIs (such as in the aquatic environment) or acute
exposure to higher levels (such as with wildlife ingestion of improperly discarded waste). Reducing API entry
to the environment has relied solely on conventional end-of-pipe pollution control measures such as wastewater
treatment and take-back collections of leftover, unwanted drugs (to prevent disposal by flushing to sewers). An
exclusive focus on these conventional approaches has ignored the root sources of the problem and may have
served to retard progress in minimizing the environmental footprint of the healthcare industry. Potentially
more effective and less-costly upstream pollution prevention approaches have long been considered imprudent,
as they usually involve the modification of long-established norms in the practice of clinical prescribing. The
first pollution prevention measure to be proposed as feasible (reducing the dose or usage of certain select
medications) is followed here by an examination of another possible approach — one that would rely on the
excretion profiles of APIs. These two approaches combined could be termed eco-directed sustainable prescribing
(EDSP) and may hold the potential for achieving the largest reductions in API entry to the environment — largely
by guiding prescribers' decisions regarding drug selection. EDSP could reduce API entry to the environment by
minimizing the need for disposal (as a consequence of avoiding leftover, unwanted medications) and reducing
the excretion of unmetabolized APIs (by preferentially prescribing APIs that are more extensively metabolized).
The potential utility of the Biopharmaceutics Drug Disposition Classification System (BDDCS) is examined for the
first time as a guide for API prescribing decisions by revealing relative API quantities entering sewage via
excretion.
© 2014 Published by Elsevier B.V.
1. Introduction
The practice of health care (the use of prescribed medications in
particular) can have a broad spectrum of potential adverse health and
economic consequences for both the environment and humans. Continuing to emerge is an understanding of the complex network of interconnected routes (Daughton, 2008; see Fig. 1 therein, also available:
⁎ Tel.: +1 702 798 2207.
E-mail address: [email protected].
http://dx.doi.org/10.1016/j.scitotenv.2014.06.013
0048-9697/© 2014 Published by Elsevier B.V.
http://www.epa.gov/nerlesd1/bios/daughton/drug-lifecycle.pdf)
that play active roles in the release to the environment of active
pharmaceutical ingredients (APIs1) from the intended use and misuse
of medications. These routes are especially important with respect to
1
Abbreviations — API: active pharmaceutical ingredient; BDDCS: Biopharmaceutics
Drug Disposition Classification System; CAFO: confined animal feeding operation; EDSP:
eco-directed sustainable prescribing; LOD: limit of detection; MEOC: Matthew Effect
Orphaned Chemical; MQL: method quantitation limit; OTC: over the counter; PBT: persistent, bioaccumulative, and toxic; PK: pharmacokinetics.
API portion excreted as reversible conjugates —>
C.G. Daughton / Science of the Total Environment 493 (2014) 392–404
393
MAXIMAL
Intermediate
(BDDCS Class IV APIs?)
Minimal
Intermediate
(BDDCS Class I APIs?)
API portion excreted unchanged as parent—>
Fig. 1. Environmental loadings of APIs as a function of excretion and reversible conjugation. See Supplemental Table S-3 for a list of example APIs (and supporting references) that have
shown “negative removals” during sewage treatment — often, perhaps, as a result of deconjugation. The possible predictive utility of the BDDCS is also indicated for Class I and Class IV APIs.
the aquatic environment [where many APIs have become ubiquitous
trace contaminants — continuously present in many waters and displaying a pseudopersistence (Daughton, 2002, 2003; Mackay et al., in
press); also see extensive list of references cited in Supplementary
Tables S-1 and S-2] as well as for both the escalating defacto reuse of
water (Rice et al., 2013) and the growing need for planned wastewater
recycling, especially for potable use (Debroux et al., 2012). The potential
for adverse impacts derives from two major routes: (1) the excretion of
unmetabolized residues of APIs (as well as their active metabolites and
“masked” derivatives such as metabolic reversible conjugates — the
parent API linked to certain endogenous biomolecules) and (2) the
accumulation of unwanted, leftover medications, whose safe and prudent
disposal is often an onerous task for the consumer and rarely performed
properly (Daughton, 2010a).
In general, excretion of API residues is the major route to the environment (especially for the aquatic domain), with adverse effects in
the aquatic environment now known to be possible at extremely low
API exposure levels. In contrast, the major concern regarding humans
is non-therapeutic exposure and self-exposure to diverted leftovers
via accidental, incidental, unintentional, or purposeful consumption —
primarily via ingestion or dermal pathways (Bond et al., 2012; Budnitz
and Salis, 2011; Burghardt et al., 2013; Daughton, 2010a). Morbidity
and mortality among infants, toddlers, teens, and the elderly (from
unintended exposure or non-medical self-exposure to diverted drugs,
both of which are exacerbated by the incidence of leftovers) are well
documented and largely preventable or avoidable. Mortality is especially
notable and discouraging since it is often preventable. Additional routes
for the entry of drug residues to the environment are bathing and dermal
transfer. These routes could be more important than excretion for select
drugs that are formulated primarily into topical preparations (such
as high-content creams and transdermal devices) and for APIs that are
extensively excreted via sweat; these routes may play significant roles
in human bystander exposure. Bathing can transfer residues to sewers
and ambient waters, while dermal contact may transfer significant residues to surrounding surfaces or directly to other people (Daughton and
Ruhoy, 2009).
Historically, problems regarding chemical contaminants in the
environment — especially those where sewage plays the major role —
have been addressed with pollution control measures. End-of-pipe
treatment is the long-established norm. Recognition has grown
over the last decade, however, that myriad numbers of trace-level
“emerging” contaminants (such as APIs) comprise the majority of the
synthetic chemicals that remain in treated sewage, even with advanced
treatment. Continual advancements needed for engineered treatment
technologies capable of removing ever-lower levels of trace contaminants from solutions are resource intensive, and limits probably exist
with regard to their economic sustainability (Jones et al., 2005).
Since the 1990s, various means of conventional and more advanced pollution control continue to be examined for reducing the
ultimate entry of APIs to the aquatic environment, especially via
treated sewage (e.g., Coday et al., 2014; Luo et al., 2014). But a singular focus on resource-intensive (and not fully effective) end-of-pipe
approaches [such as improved treatment technologies for wastewater
and drinking water, and “take-back” programs for collection of unwanted leftover medications to avert their disposal by flushing to sewers
(e.g., Glassmeyer et al., 2009)] ignores the root origins of the problem
and may actually serve to retard meaningful progress in minimizing
the ecological and chemical footprints of the healthcare industry.
In contrast, pollution prevention is a major unexplored approach for
minimizing the impact of healthcare on the environment. Preventative
measures would target the root factors that promote or facilitate the
release of APIs to the environment. The most important routes for the
release of APIs to the environment are excretion (unmetabolized API
or active metabolites), bathing (topical APIs and sweat), and imprudent
disposal of leftover, unwanted medications (especially to sewers).
The key up-stream processes that dictate the scope and magnitude of
excretion are the regulations, guidelines, behaviors, and customs surrounding the practice of prescribing and ultimate use, along with the
associated activities of dispensing as influenced by the administration
of healthcare and the insurance industry (Daughton, 2013; Ruhoy and
Daughton, 2008).
1.1. Background
The practice of health care involves the widespread use of roughly
2500 distinct active pharmaceutical ingredients (APIs) in the US
(roughly 4000 worldwide) formulated into tens of thousands of commercial pharmaceutical preparations (Daughton, 2013). The intended
ultimate use of these APIs — some of which can elicit biological effects
at the nanomolar level and below — often results in the excretion
394
C.G. Daughton / Science of the Total Environment 493 (2014) 392–404
(primarily via urine or feces, and secondarily via sweat) of unmetabolized APIs or bioactive metabolites. APIs can differ dramatically with
regard to the extent of excreted dose — from practically nil to nearly
complete; but there are few drugs for which metabolism (and excretion) are intermediate — i.e., between 30% and 70% (Benet et al.,
2011). These two extremes encompass most APIs, which are either
extensively metabolized or extensively excreted unchanged. Furthermore, portions of many oral dose forms are never absorbed
systemically — a result of being excreted immediately and directly via
the feces; this mechanism clearly serves to maximize the percentage
excreted unchanged.
Excreted APIs enter the aquatic environment by way of both treated
and untreated (raw) sewage; APIs in raw sewage enter unabated into
surface and ground waters not just by wet-weather runoff and illegal
discharges, but also by contributions from numerous point sources
from defective sewer connections (Baum et al., 2013). Leftover, unwanted medications are also often disposed into sewers. Some APIs
are formulated for external use (high-content topical drugs); some
of these APIs have exclusive topical use (they are not administered
systemically). For these APIs, bathing is a major route of entry to the
environment (Daughton and Ruhoy, 2009). Both excretion and the
need for disposal are partly driven by imprudent, unnecessary, or excessive prescribing, misuse, and overconsumption — all major problems in
healthcare and the ones with many, complex causes (Daughton and
Ruhoy, 2011).
Significantly, current approaches directed at reducing API levels in
the environment have focused solely on pollution control — particularly
improved wastewater treatment and take-back collection of unused
consumer medications. These are end-of-pipe approaches, which for
decades have been the hallmarks for controlling chemical contamination of the environment. These are not, however, approaches that can
be relied upon to facilitate the sustainable use of medications. To the
contrary, an argument may exist that pollution control measures
might work counter to sustainability by deflecting the ongoing dialog
surrounding drug residues in the environment away from possibly
more effective measures addressing pollution prevention. The absence
of a focus on pollution prevention fosters continued, unfettered prescribing and use of unnecessary drugs, for excessive durations, and
often in excessive doses.
Many of the aspects of a drug's life cycle that have been identified as
possible targets for optimizing to reduce API entry to the environment
involve alterations to prescribing and dispensing practices. Some of
these practices have already been undergoing examination for other
purposes, such as improving patient adherence or compliance with
medication regimens to reduce adverse events and improve therapeutic
outcomes (Daughton, 2010a). Other prescribing modifications include drug substitution, reducing dispensed drug quantity (especially
amounts suitable for short-term trials), easier or better-targeted delivery systems (e.g., transdermal systems), lower doses [e.g., achieved
with alternative delivery routes or personalized doses (Daughton and
Ruhoy, 2013)], dose timing (e.g., chronobiology), palatability (a factor
that can strongly influence patient compliance and thereby raise or
lower the incidence of leftovers), physician medication reviews with
patients (and prevention of unnecessary polypharmacy), more informative and clearer labeling (which can directly promote patient compliance), elimination of unnecessary repeat prescriptions (especially
automatic refills), improved coordination among prescriber, dispenser,
and patient, and alternative treatments (exercise, physical therapy,
diet, etc.). Numerous other approaches involve design of API chemical
structure, drug formulation, and packaging. While many of the modifications to prescribing practices are intended to improve patient compliance and adherence, they may also coincidentally serve to reduce
the incidence of leftovers and the subsequent need for disposal. Other
potential approaches have included consideration of pharmEcokinetic
factors [e.g., prescribing decisions partly based on selection of drugs
having lower half-lives in the environment or reduced propensity to
undergo bioaccumulation (Deblonde and Hartemann, 2013; Stockholm
County Council, 2012)]. The spectrum of potential options for gaining
better alignment with sustainability is clearly vast.
To minimize the potential for APIs to enter sewers in the first place, a
wide array of measures designed to reduce or eliminate drug wastage
have been under consideration or evaluation; these are partly designed
to reduce the disposal of leftover medications in sewers. Up to now, the
major approach widely assumed to lessen the occurrence of APIs in the
aquatic environment has been the implementation of federal, state, and
local guidelines for discouraging sewer disposal of leftover and unwanted drugs; this rationale persists despite the lack of evidence that sewer
disposal contributes significant quantities of most APIs to the quantities
already unavoidably entering sewage via excretion (Daughton, 2010a).
Potentially effective and less-costly upstream pollution prevention
approaches have long been considered imprudent and impractical
simply because they might conflict with long-accepted prescribing
guidelines, norms, and tenets. But these are often influenced by behaviors, customs, attitudes, and traditions — of prescribers and patients
alike. All these combined have contributed to an unfounded fear of
jeopardizing the quality of delivered health care if prescribing guidelines are altered. Long-deemed infeasible has been the optimization of
the therapeutic use of medications for preventing pollution at its source.
This stance, however, has been shown to be unfounded in at least
one instance, where some select drugs can be prescribed at off-label
doses considerably lower but still prudent and efficacious; such lower
doses could reduce wastewater loadings from excretion. Moreover,
lower-doses hold the potential to also avoid the subsequent need for
disposal of leftovers that would otherwise be generated as a result of
patient non-compliance caused by adverse effects from higher doses
(Daughton and Ruhoy, 2013). Many drugs are prescribed to segments
of the population at doses that are unnecessarily or imprudently high.
Imprudent drug prescribing and ultimate use are major aspects of escalating health care costs, which overall compose an unsustainable 17.6%
of GDP (Curfman et al., 2013). Furthermore, by reducing the incidence
of leftovers via lower doses, a concomitant reduction could result in
drug diversion, abuse, and unintended poisonings (Daughton and
Ruhoy, 2013). These are all major problems in the U.S. and a primary
concern for the White House Office of National Drug Control Policy
(ONDCP, 2013). To date, however, dose-reduction has been the only
proposed approach for directly reducing the primary pathway (excretion)
for API release to the environment, as well as for reducing the incidence
of leftovers and the consequent need for their disposal. This proposed
approach also suggests that patients and prescribers can consider more
prudent medications and regimens; reducing the overuse and imprudent
use of antibiotics is one example (Daughton, 2010a; Daughton and Ruhoy,
2013).
This first proposed approach to pollution prevention (lower-dose
prescribing, see: Daughton and Ruhoy, 2013) is now followed here by
an examination of a complementary but potentially more expansive approach for controlling the major route of API entry to the environment —
excretion — which has escaped concerted attention as a target for
control. Never before considered is a pollution prevention approach
designed around the excretion profiles of APIs — favoring those that
are more extensively metabolized to benign end products versus those
known to be extensively excreted unchanged as the parent API or as
reversible metabolic conjugates. Presented here is an examination of a
concept for formally accommodating API pharmacokinetics (namely,
API excretion parameters) in the decision process surrounding the practice of clinical prescribing. Such eco-directed sustainable prescribing
(EDSP) could prove central to the advancement of a sustainable
healthcare system while protecting the environment — treating the
patient and the environment as an integral, interconnected whole.
Excretion profiles could also identify those APIs on the other end of
the spectrum — those that are extensively excreted unchanged. For
these APIs, the continued disposal of any unwanted leftovers to sewers
could perhaps be justified on the basis that the quantities of excreted
C.G. Daughton / Science of the Total Environment 493 (2014) 392–404
395
residues may far surpass the incremental contributions from disposal.
For these drugs as unwanted waste, their continued immediate disposal
to sewers could be favored also because flushing remains the most
effective practice uniformly accessible to consumers for ensuring that
certain drugs are not diverted for abuse and for preventing unintended
poisonings in humans and pets; leftover medications continue to be one
of the leading causes of accidental mortality in children (Bond et al.,
2012; Budnitz and Salis, 2011; Burghardt et al., 2013; Daughton,
2010a). Those APIs that are extensively metabolized could then be
targeted as priorities for finding alternative pollution prevention approaches for disposal, since the contribution of their residues via other
pathways (such as flushing leftovers into sewers) would pose a greater
probability of adding significant portions to overall environmental
loadings.
Examined here is the feasibility of factoring API excretion profiles
into the decision process for prescribing and dispensing in order to
optimize the selection of drugs posing minimal potential for environmental impact via excretion. Within given therapeutic classes, particular APIs may exist with more favorable metabolic profiles — those
resulting in less excretion of bioactive residues. With an understanding
of an API's pharmacokinetics (PK) that is more comprehensive than currently available (such as the routine PK data compiled in PK databases
or provided in patient package insert documentation), an API within a
given therapeutic class could be selected partly on the basis of reduced
excretion. This approach could most easily be first implemented for
those therapeutic groups where the APIs display minimal differences
in therapeutic effectiveness. Certain drug classes (especially cytotoxic
chemotherapeutics) may not be amenable to this approach; the best
control measure for such highly toxic drugs may simply be the prevention of urine and feces from entering sewers.
EDSP. A major challenge in trying to catalyze change in society's
relationship with pharmaceuticals is the sheer number of stakeholders
concerned with the many aspects of the lifecycle of drugs — spanning
from the point of manufacture and extending to prescribing, dispensing,
ultimate usage, storage, diversion, disposal, and treatment (Daughton,
2008; see Fig. 1 therein, also available: http://www.epa.gov/nerlesd1/
bios/daughton/drug-lifecycle.pdf).
A notable aspect of EDSP would be that improvements to the practice of conservative prescribing that are aimed at either reducing API
excretion or the incidence of medication leftovers will at the same
time also serve to improve aspects of healthcare and public safety. As
previously argued for reduced doses (Daughton and Ruhoy, 2013),
EDSP could have the same far-reaching collateral benefits, including
reduced healthcare costs (by reducing dose and reducing medication
waste), improved patient therapeutic outcomes (by reducing adverse
events, thereby improving patient adherence, which in turn dictates in
part what portion of a course of medication remains unused and thereby eventually requires disposal), and reduced morbidity and mortality
from accidental poisonings caused by improperly stored or disposed
medications. Leftover, unwanted medications are overt symptoms
and direct measures of numerous inefficiencies and imprudence in the
conduct and administration of healthcare. They directly reflect wasted
resources (in terms of physician time and consumer expense), lost
opportunities to achieve therapeutic outcomes (when leftovers are
generated as a result of patient non-compliance or non-adherence),
and pose significant but avoidable hazards to public safety and health
(via diversion, abuse, and unintended poisonings) as well as to wildlife
(Daughton and Ruhoy, 2011).
1.2. Objectives
Despite the ready availability of limited PK data for drugs, comprehensive PK data (sufficient to estimate API levels that would reach the
environment after metabolism) can be surprisingly difficult to locate;
the pharmacokinetics for many drugs are still not even sufficiently
understood. This is because PK data needed for clinical trials and drug
registration purposes do not need to account for the portion of a dose
that passes directly through the gut unabsorbed and unmetabolized
(sometimes exceeding the majority of a dose) or for reversible
metabolic conjugates (those that can undergo deglucuronidation, via
microbial or abiotic hydrolysis) versus total conjugates, which include
non-reversible conjugates formed from phase I metabolites; conjugates
of phase I metabolites do not yield the parent API upon hydrolysis
(Hermening et al., 2000). The data cited in studies involving predicted
environmental concentrations (PECs) for APIs often simply state that
an API is “extensively metabolized” or “extensively excreted” and are
insufficient to rule out whether the API has potential for occurrence in
the environment via excretion.
The best available PK studies are those that strive to achieve
stoichiometric mass balance around the parent API and all identified
excreted metabolites (including all forms of metabolic conjugates)
and unchanged parent API; these comprehensive studies usually involve mass-balance around radiolabeled APIs (White et al., 2013). But
even then, these types of comprehensive studies involve few subjects.
Within a population, many factors can dramatically modulate pharmacokinetics, resulting in enhanced or reduced excretion of parent API.
Examples among numerous others include: dose, dose formulation
(e.g., extended release; influence of excipients on absorption), duration
of treatment, chronobiology, genetic polymorphisms (e.g., extensive
versus poor metabolizers), gut microbiota, stress, exercise, diet, gender,
age, physiology (especially intestinal physiology affecting motility and
pH), health status (especially bowel disorders), and polypharmacy
(e.g., drug–drug interactions, which can profoundly influence phase I
metabolism, for example). Numerous drug-specific factors also influence the PK of APIs, notably including dissolution (e.g., Charkoftaki
et al., 2010; Jamei et al., 2009; Macheras et al., 2013; McConnell et al.,
This project originally set out to provide a foundation for understanding the preventative measures that could be implemented for
circumventing the entry of APIs to the environment. This could be accomplished first by reducing doses for certain APIs (when feasible and
prudent) (Daughton and Ruhoy, 2013) and, now here in this article,
by selecting medications whose APIs have more favorable excretion
profiles. These two pollution prevention approaches combined could
be called eco-directed sustainable prescribing (EDSP). The premise
is that EDSP holds the potential for achieving the largest reductions
in aquatic levels of API contaminants by reducing the major source
(excretion) as well as a secondary source (disposal of leftovers to
sewers). And at the same time, EDSP holds promise for improving
the efficacy or healthcare while also reducing costs.
The objective in this paper is to determine what type of PK data
would be needed (and how these data could be most readily obtained)
to help in selecting APIs for two major purposes: (1) those APIs whose
excretion is minimal (and could therefore be classified as having
lower potential for environmental impact — when used as prescribed),
and (2) those APIs whose excretion is maximal and therefore disposal of
leftovers to sewers might have minimal comparative impact on the
aquatic environment (versus the quantities normally excreted); for
the latter group of APIs, disposal to sewers could possibly continue as
a recommended practice when human safety and health are a priority
(e.g., when drug diversion exacerbates human morbidity and mortality)
(see list of APIs at: USFDA, 2009).
Ultimate objectives are to foster a better understanding among the
healthcare communities as to how the use of pharmaceuticals impacts
the environment — and indirectly may impact the general public via
a number of routes — including de facto recycled drinking water
(Daughton, 2010b; Debroux et al., 2012; Rice et al., 2013) — and to
facilitate or catalyze discussion and further work among the many
stakeholders regarding pollution prevention and environmental stewardship and how these impacts could be significantly reduced with
2. Materials, methods, and approach
396
C.G. Daughton / Science of the Total Environment 493 (2014) 392–404
2008); the critical role of dissolution is shown by rifaximin, which is
directly excreted, completely unchanged — almost exclusively in feces
(Karanje et al., 2013). All of these variables can lead to considerable
variance among individuals and across populations. This consequently imparts great uncertainty to predicting API input to sewers
via excretion.
The original intent of this project was to compile comprehensive PK
data on a wide spectrum of APIs. These APIs would be ranked according
to the propensity of the parent API to be excreted. This ranking could
then be used as an additional factor in guiding prescribing decisions —
with the intent of reducing the overall loadings of APIs via sewers. For
example, this excretion footprint could essentially serve as a fourth
criterion, in addition to the three currently used for the Stockholm
“Wise List” model of “Environmentally Classified Pharmaceuticals”,
created for the Stockholm City Council (Wennmalm and Gunnarsson,
2010). This represents the first and currently only formal system
for classifying medications with respect to their potential for environmental impact. This system has been implemented in the form of
“eco-labeling” and was designed to assist the prescribing process by
considering the potential for environmental impact. A major limitation,
however, is that the Stockholm criteria only comprise the three conventional factors (termed PBT) long-used in prioritizing chemicals for
potential environmental harm: persistence (e.g., reflected by biodegradability), bioaccumulation (e.g., proxied by octanol-water partition
coefficients), and aquatic toxicity. Importantly, however, these three
factors only come into play if and when an API enters the environment.
A more realistic approach needs to consider the potential for an API
to gain entry to the environment to begin with. After all, an API with
unfavorable PBT characteristics may actually have eco-friendly PK properties, imparting it with little potential to enter the environment — even
if consumed by a large segment of the population. EDSP would add
a fourth dimension to the Wise List — one that factors in PK excretion
profiles — primarily the propensity for excretion of structurally unchanged APIs, reversible conjugates, and eco-toxic metabolites.
Quickly becoming apparent, however, is the difficulty in mining
comprehensive PK excretion data for numerous APIs from the primary
literature. The available data rarely are sufficient to account for reversible conjugates, which can serve as a major source of an API in the environment (beginning during transit of waste to an STP). This can be
readily seen with studies of API levels in STPs where the concentrations
in effluents are often significantly higher than in the influents (see discussion in Section 2.6: “Limitations to data — Factors influencing environmental occurrence and its measurement”). This also means that
the excretion data used in published models to estimate API excretion
to sewers are unable to accurately account for reversible conjugates
(Lienert et al., 2007; Yan et al., 2014).
2.1. Proxy measure for API excretion: the BDDCS
Instead of an approach involving mining PK data from the literature,
an alternative measure was evaluated in the study reported for
the first time here. This approach makes use of what is called the
Biopharmaceutics Drug Disposition Classification System (BDDCS) —
an existing system used in the pharmaceutical industry for predicting
various pharmacokinetic properties of APIs.
A discussion on the background and foundation of the BDDCS is
beyond the scope of the work presented here but it is available from a
number of articles (Benet, 2013; Benet et al., 2011; Custodio et al.,
2008; Pham-The et al., 2013); the BDDCS serves as an extension of the
predecessor work on the Biopharmaceutics Classification System
(BCS) (Wu and Benet, 2005). Both systems attempt to classify APIs
according to two major parameters. The BDDCS uses solubility and
intestinal permeability — yielding four combinations of high and low
(Classes I through IV); additional but small classes (e.g., Classes 0 and V)
comprise a select few APIs whose PKs are extremely sensitive to pH
profiles (e.g., amphetamine) or that display facile and ready degradation
in the gut. It is important to note that class assignments for certain APIs
are provisional and are subject to revision (Pham-The et al., 2013).
The APIs from only two of the four BDDCS classes were selected
for the study reported here because they most likely represented two
extremes with respect to the propensity of an API to be extensively metabolized (Class I) or to be extensively excreted unchanged (Class IV).
This study did not evaluate Class II or Class III APIs, which probably
would represent intermediate propensities. BDDCS Class I currently
represents 40% of marketed drugs and 18% of new molecular entities
(NMEs), while Class IV only represents 6% of marketed drugs and
NMEs (Benet, 2013). This is the reason for the discrepancy in the number of APIs selected from these two classes.
On paper, the use of PK data for predicting the excretion of unchanged API should be a useful tool for predicting API entry to the environment. A host of factors would need to be considered, however,
in evaluating the excretion efficiency of an API. Mining such data for
each AP would be a time consuming task — made rather futile because
the data may not be representative of reality (for any number of the reasons summarized earlier). Consideration of just one variable illustrates
the complexity of the proposition. Consider carbamazepine, which is
one of the most frequently detected APIs in the environment — despite
the fact that it is extensively metabolized via the liver, with conjugation
primarily of phase I metabolites. Carbamazepine might be gaining entry
to sewers not because of any quirks of metabolism, but rather as a result
of its slow, erratic, and highly variable rate of dissolution in the gut — a
result of its poor aqueous solubility (a major limitation for BDDCS Class
II APIs) (Hardikar et al., 2013). This can lead to substantial undissolved
quantities passing directly through the gut — evading uptake during
gastrointestinal transit. Poor dissolution of dose forms was proposed
in 2001 as a factor promoting the entry of at least some APIs to the
environment (Daughton, 2001). The compounding effects of meals
(especially lipids) and non-homogeneous mixing within the gut add
yet more variability (e.g., Schiller et al., 2005). Just by consideration of
the unpredictable variability in excretion introduced by the dissolution
of a drug during its transit through the gut, it becomes clear that the use
of PK for predicting an APIs entry to the environment would be vulnerable to considerable error.
With this as a driver, the BDDCS was examined as a proxy measure
for the relative extent of excretion of APIs unchanged. As a proxy measure for excretion, the BDDCS may not be as rigorous as compiling comprehensive PK data from the published literature, but it offers a number
of advantages — the primary ones being its simplicity, ready accessibility, and recognition within the drug development community. The current study examined whether the published environmental occurrence
levels of Class I APIs (measured in various environmental compartments
but with emphasis on sewage and surface waters) trended lower than
the levels for the Class IV APIs. That is, did the APIs belonging to the
extensively metabolized group (BDDCS Class I) tend to have associated
environmental monitoring levels that were clearly lower than the
APIs in the group that was extensively excreted unchanged (BDDCS
Class IV). The use of empirical field-monitoring data essentially accounted
for the numerous variables involved with an API's entry to and transit
through the aqueous environments of sewage and ambient waters.
2.2. Unanticipated outcome
A major collateral outcome resulted from this study in the course of
mining the published environmental occurrence data for the APIs that
were selected from the two BDDCS classes as presented in Benet et al.
(2011). The result (compiled in Supplemental Tables S-1 and S-2)
represents one of the larger and more comprehensive snapshots of
the published data for the environmental occurrence of APIs; a number
of prior efforts have also cataloged occurrence data for various APIs
(e.g., Barnes et al., 2008; Daneshvar, 2012; Deo, 2014; Deo and
Halden, 2013; Focazio et al., 2008; Hughes et al., 2013; Kolpin et al.,
2002; Verlicchi et al., 2012; Williams and Cook, 2007; Zhou et al.,
C.G. Daughton / Science of the Total Environment 493 (2014) 392–404
397
2009). The data compiled in this current examination includes not
just data of presence and data of absence, but in some respects more importantly it reveals those APIs for which data are completely lacking
(absence of data). The published literature was examined for 374 APIs
and involved the mining of data from over 500 articles (primarily
from journals, book chapters, reports, and dissertations). Summaries
of the data compiled in Tables S-1 and S-2 are provided in Section 3
(Results and conclusions) within Tables 1 and 2.
The importance of negative data and absence of data should not be
underestimated. Consistent data of absence tells us which APIs might
be lower priorities for future monitoring or what we might be able to
ignore, thereby conserving resources. In contrast, the absence of data
tells us what we might need to begin targeting for examination. With
respect to the therapeutic use of drugs, data of absence in the environment (in conjunction with drug usage statistics and knowledge of
metabolites of potential environmental concern) might tell us which
APIs could continue to be used therapeutically with minimal environmental impact (although the potential for human poisoning from
diverted drugs may still exist).
Table 2
The APIs from BDDCS Class IV APIs (total of 52) for which environmental occurrence data
seemed to exceed a threshold level of 1 μg/L in waters or 1 mg/kg in solids (for complete
data, see Supplemental Table S-2).
Table 1
The APIs from BDDCS Class I APIs (total of 322) for which environmental occurrence data
seemed to exceed a threshold level of 1 μg/L in waters (or 1 mg/kg in solids) (for complete
data, see Supplemental Table S-1).
2.3. Literature search process
Abundant occurrence data (57 APIs total in this group)
Acebutolol hydrochloride
Alprazolam
Aminophenazone
Amitriptyline (N5 μg/L; max 11.1 μg/L)
Bromazepam (N5 μg/L; max 15.5 μg/L)
Butalbital (N5 μg/L; max 5.3 μg/L)
Chloramphenicol (N5 μg/L; max 40 μg/L)
Cyclophosphamide (N5 μg/L; max 13.1 μg/L)
Diazepam
Diclofenac
Diltiazem
Diphenhydramine
Enalapril (N5 μg/L; max 10 μg/L)
Ethinylestradiol
Hydroxyzine
Ketamine
Meprobamate
Metoprolol
Metronidazole
Minocycline (N1 mg/kg)
Omeprazole
Phenobarbital
Risperidone
Sertraline
Temazepam
Tramadol (N5 μg/L; max 86 μg/L)
Venlafaxine
Limited occurrence data (41 APIs total in this group)
Escitalopram (N5 μg/L; max 32.2 μg/L)
Ramipril (N5 μg/L; max 5.4 μg/L)
Secobarbital (N5 μg/L; max 30 μg/L)
Zolpidem (=5 μg/L)
Zopiclone (=1 mg/kg)
Paucity of occurrence data (224 APIs total in this group)
Butabarbital
Chlordiazepoxide (N5 μg/L; max 6 μg/L)
Clorazepate (N5 μg/L; max 6.2 μg/L)
Doxorubicin (N1 mg/kg; max 5.6 mg/kg)
Indapamide (N5 μg/L; max 15.4 μg/L)
Linezolid (N5 μg/L; max 6 μg/L)
Levodopa
Phenylephrine
Valacyclovir (N5 μg/L; max 5.7 μg/L)
Valproic acid (N5 μg/L; max 9.3 mg/kg)
Zidovudine (N5 μg/L; max 9 μg/L)
Abundant occurrence data (13 APIs total in this group)
Ciprofloxacin (max 3.5 mg/kg)
Enoxacin (max 1.3 μg/L)
Erythromycin stearate (max 1 mg/kg)
Fleroxacin (max 1.84 mg/kg)
Furosemide (N1 μg/L; max 3.2–3.8 μg/L)
Norfloxacin (max 5.6 mg/kg)
Penicillin V (max 13.8 μg/L)
Roxithromycin (N1 μg/L; max 5 mg/kg)
Sulfamethizole (max 5.2 μg/L)
Valsartan (max N5 μg/L)
Limited occurrence data (8 APIs total in this group)
Acyclovir (max 1.76–2.4 μg/L)
Chlorothiazide (max 4.5–8.9 μg/L)
Chlorthalidone (max 20.1 μg/g)
Eprosartan (max 6.8 μg/L)
Paucity of occurrence data (31 APIs total in this group)
No data were available for 22 of the APIs in this group.
Of the few data available, none exceeded the threshold levels.
The primary source of data that was used to mine API environmental
occurrence levels (or to verify the absence of data) is a bibliographic
database maintained at the US EPA. The scope and coverage of this
database are described here: http://www.epa.gov/ppcp/pdf/Synopsisof-PPCPs.pdf. This database is one of the largest available that is devoted
exclusively to all of the many and complex issues surrounding the interface between pharmaceuticals and the environment (Daughton and
Scuderi, 2014); as of this report, this database contained over 18,500 records, including archival journal articles (published as well as in-press),
book chapters, dissertations, reports, web pages, and the gray literature,
among others; coverage dates back primarily to the 1980s, which
coincides with the advent of concerted study of pharmaceuticals in
the environment. All documents added to the database (compiled in
EndNote X7, Thomson Reuters) were examined to ensure that their
contents were digitized; when the main bodies of documents comprised scanned images, they were digitized (using Adobe Acrobat X
Professional). Over 96% of the journal articles had complete digitized
reprints allowing fast, full-text searching. This bibliographic database
has been updated and curated on nearly a daily basis since 2008. Its
articles are mined from commercial and public on-line databases,
none of which provides comprehensive coverage on its own. These
databases include ScienceDirect, American Chemical Society, Wiley,
Springer, Taylor & Francis, Google Scholar, MedLine/PubMed, and the
web itself. Hits from primary searches were expanded with reverse
and forward citation analysis to accelerate location of additional relevant references and as a quality check on completeness. This database
facilitates fast, full-text Boolean keyword searches. Most importantly,
however, since the database content has already been triaged and curated for relevant articles, the searches avoid the major problem of numerous extraneous hits, which are inevitable whenever searching for data
regarding pharmaceuticals within non-curated databases. The search
strategy was not capable of locating articles where the spelling of the
API search term was highly unusual (e.g., some non-English language
spellings), nor could it locate references where the API spelling was
consistently incorrect.
2.4. Caveats and comments regarding literature searching
Examination of the published literature surrounding the environmental occurrence of APIs reveals two distinct groups: (1) those APIs
that have been specifically targeted for detection or quantitation in
398
C.G. Daughton / Science of the Total Environment 493 (2014) 392–404
any number of different matrices or environmental compartments, and
(2) those APIs that have never (or rarely) been targeted for any type of
environmental monitoring. The occurrence levels for APIs in the first
group span the gamut from levels below the limits of analytical detection or quantitation, to the commonly reported levels encompassing
the ppt–ppb range, and the less-common levels that span the ppb–
ppm range. This first group therefore comprises both positive and
negative data (i.e., data of presence and data of absence). The second
group comprises APIs with an absence of data. These APIs have escaped
targeted analysis in the environment for any number of reasons, ranging
from the lack of suitable analytical methodologies to an outright lack
of attention. Some of these APIs lacking data of occurrence may belong
to a group referred to as Matthew Effect Orphaned Chemicals
(MEOCs), as discussed in Daughton (2014), and therefore possibly
merit future scrutiny.
For any individual API in the first group, the published occurrence
data often do not cluster in clear or defined ranges. Instead, the occurrence data for an API often span the spectrum from non-detection to
levels exceeding 1 μg/L in waters or 1 mg/kg in solids. This makes it difficult to generalize or to rank APIs according to their prevalence in the
environment — whether by frequency of occurrence, geospatial distribution, environmental compartment, or especially concentration levels.
Moreover, the reported levels among APIs are not intercomparable because limits of detection (LODs) or method quantitation limits (MQLs)
can differ by one or more orders of magnitude. One consequence
when comparing levels among APIs, for example, is that an API with
abundant data of absence could actually occur at levels higher than an
API that has a lower LOD and abundant positive data (albeit low levels).
In this current project, concentration data were mined from examination of publications covering environmental matrices — with a primary focus on waters (especially sewage and natural surface waters);
groundwaters, source and finished drinking waters, and biota were of
less interest because of the increased probability that the API levels
had been yet further diminished by any number of transformation processes. Individual searches were performed for nearly every API compiled in the BDDCS evaluation that was performed by Benet et al.
(2011). A small group of APIs (21 of the 346) from BDDCS Category I
were excluded from evaluation because they have little toxicological
relevance in the environment or they have major alternative contributory sources beyond that from bona fide human consumption of pharmaceuticals, such as from: endogenous biosynthesis (e.g., many of
the estrogens, hydrocortisone, melatonin, vasopressin), food sources
(caffeine, theophylline, niacin, cholecalciferol), illicit drug consumption
(e.g., morphine, cocaine), widespread abuse (e.g., ethanol, nicotine), or
domestic animal use (e.g., ivermectin). Occurrence data for these few
APIs may therefore not reflect human excretion from ultimate therapeutic use.
The published occurrence data (both positive and negative) were organized into three somewhat subjective groups (see summaries
below): APIs with: (1) abundant occurrence data, (2) limited data,
and (3) paucity of data. These data can include non-detects (data of absence). The compiled data emphasized API occurrence in STPs and surface waters, while attempting to exclude data from locations possibly
biased with contributions from hospitals or other healthcare facilities,
manufacturing facilities, and confined animal feeding operations
(CAFOs). No attempt was made to convert and standardize the reported
units of concentration, such as ng/mL versus μg/L, or between ng/g, μg/kg,
and mg/kg. The published literature was searched up through 8 May 2014
using the bibliographic database of Daughton and Scuderi (2014).
2.5. Three groups of API occurrence data
2.5.1. Abundant occurrence data
API is frequently detected in a wide range of matrices; levels reported by isolated studies are infrequently appreciable (greater than 1 μg/L
or 1 mg/kg) but can also be low — probably a function of the quantity of
drug locally prescribed or consumed. Numerous additional supporting
references exist beyond the few examples cited in Supplemental
Tables S-1 and S-2, which were selected primarily from the more recent
literature. Asterisks in the column “Reported occurrence data” denote
that published occurrence data supports an API's presence at substantial
levels (i.e., levels in STPs exceeding 1 μg/L, or levels in sludges or sediments exceeding 1 mg/kg or 1 μg/g).
2.5.2. Limited occurrence data
API has been much less frequently targeted for monitoring and usually only in a limited number of matrices (primarily limited to STP
wastewaters — raw influent or treated effluent). In contrast to the references cited for the “Abundant occurrence data” group, the references
cited for “Limited occurrence data” are comprehensive, representing
all that could be located in the published literature.
2.5.3. Paucity of occurrence data
A paucity of data does not imply that occurrence levels are low or
below LODs, but rather that there have been at most very few studies
that have targeted the API for monitoring (or multiple studies might
exist but they are from the same authors); one or two isolated studies
might report comparatively low or high levels but no sense of representativeness can be gained. With the exception sometimes of isolated
reports, essentially no published occurrence data could be located (including data of absence). The cited references represent a comprehensive examination of the published literature. Many of these APIs are
possibly Matthew Effect Orphaned Chemicals (MEOCs) (Daughton,
2014), and may therefore deserve attention as targets for future monitoring efforts.
2.6. Limitations to data — factors influencing environmental occurrence
and its measurement
There are numerous complexities and limitations in interpreting the
environmental occurrence data for APIs. Although these are important
to understand, this section can be skipped by the reader without
compromising an understanding of the subsequent sections.
Many of the higher API occurrence levels captured in Tables S-1 and
S-2 were isolated reports and may have been erroneous or isolated excursions; for those in the group with abundance of data, there may have
been additional data that could have further raised the maximum levels
compiled in the tables. Note that an abundance of data does not necessarily correlate with widespread geographic occurrence, as very low
levels (e.g., fluvoxamine) or non-detection (data of absence) is also
often reported (e.g., cyclophosphamide). Other APIs may frequently
have both data of absence and data of occurrence (examples include lorazepam and omeprazole).
Some APIs may be frequently detected (and at higher levels) in STP
influent but not effluent (e.g., cortisone) and vice-versa. Apparently
higher API levels in STP effluent versus the paired influent (so-called
“negative removals”) may often result from a variety of mechanisms, including deconjugation (hydrolysis of reversible metabolic conjugates,
e.g., ketamine; budesonide). Reversible conjugates essentially serve as
“masked” forms of APIs — serving as hidden reservoirs that when hydrolyzed are converted back to the parent form. Failure to account for
reversible conjugates in predictive models can yield occurrence data
that would point to much lower than actual environmental levels.
See Supplemental Table S-3 for a listing of many examples of nonsteroidal APIs for which so-called “negative removals” have been reported (such as resulting from hydrolysis of reversible conjugates)
and Fig. 1 for the possible role of excretion in determining the environmental loadings of BDDCS Class I versus Class II.
The sewage-mediated deconjugation hypothesis emerged in the late
1990s, but the initial focus was steroids (Desbrow et al., 1998; Panter
et al., 1999; Ternes et al., 1999). Many studies have since shown the incidence of higher levels for many APIs and endogenous hormones in STP
C.G. Daughton / Science of the Total Environment 493 (2014) 392–404
effluents versus influents result from deconjugation during sewage
transit or treatment (e.g., D'Ascenzo et al., 2003; Kumar et al., 2012;
Liu and Kanjo, 2012; Verlicchi et al., 2012); see listing of references providing data for various non-steroidal APIs (Supplemental Table S-3).
Other instances of negative removals may result from the release of
parent API from suspended fecal materials or flaws in sampling design
or higher MQLs for influent than effluent (e.g., Blair et al., 2013; Gao
et al., 2012; García-Galán et al., 2012; Snyder et al., 2007; Sui et al.,
2011; van der Aa et al., 2011; Zhang et al., 2013).
Of the APIs in BDDCS Class I that have a paucity of environmental
occurrence data, a portion may qualify as MEOCs (i.e., they have simply
escaped notice), but others may have been actively ignored or overlooked for any of a wide spectrum of reasons. The occurrence data for
others may be convoluted because of contributions from multiple
sources (e.g., specific APIs originating from two or more related APIs,
such as prodrugs). The following briefly summarizes some of these
complicating factors.
Some APIs originate from their use as APIs in their own right but also
from prodrug APIs. For example, the following APIs can originate as the
major active metabolites from their respective prodrugs (shown in
parentheses): fluorouracil (capecitabine), meprobamate (carisoprodol),
prednisolone (prednisone), and primidone (phenobarbital). Some can
also originate as metabolites from APIs not specifically designed as
prodrugs. For example, clofibric acid is an active metabolite shared
among multiple fibrate prodrugs; temazepam is also a metabolite of
diazepam, and itself also yields the API oxazepam as a metabolite;
nortriptyline is also the active metabolite of amitriptyline; nortilidine
is also the active metabolite of tilitdine; oxymorphone is also the active
metabolite of oxycodone; desalkylflurazepam is the major active metabolite of flurazepam and quazepam; and desipramine is the major
active metabolite of imipramine.
Others yield related APIs as metabolic products (e.g., hydroxyzine
yields the metabolite cetirizine). Still others are inactive themselves,
serving as prodrugs for their active metabolites; as examples, the following APIs serve as the inactive prodrug esters of their respective
APIs (shown in parentheses): benazepril (benazeprilat), bopindolol
(pindolol), capecitabine (5-fluorouracil), cilazapril (cilazaprilat), enalapril (enalaprilat), imidapril (imidaprilat), olmesartan medoxomil
(olmesartan), oseltamivir (oseltamivir carboxylate), perindopril
(perindoprilat), ramipril (ramiprilat), temocapril (temocaprilat), valacyclovir (acyclovir), and valganciclovir (ganciclovir). For inactive APIs
such as these, monitoring for the prodrug active metabolites would
often be more useful than monitoring for the prodrugs themselves.
Although some of the APIs subject of this examination may be extensively excreted unchanged — and occur widely and frequently — their
measured levels are very low (some below method detection limits,
such as norgestimate) because of high potency and therefore low doses
and low manufactured quantities (e.g., norethindrone, norgestrel).
Some of these APIs have not been approved for use, have restricted
use, or have been withdrawn from the market in some countries
(e.g., benidipine, buflomedil, cerivastatin, chloral hydrate, chlordiazepoxide, dezocine, dilevalol, mianserin, rosiglitazone maleate, sibutramine, temocapril, tropisetron, urapidil, vorozole), perhaps explaining
why they have not been targeted for monitoring; others are approved
only for veterinary use in certain countries (e.g., phenylbutazone;
promazine) or are no longer manufactured (e.g., molindone). Note, however, that even though some drugs have been removed from the market,
they may still experience use. The widespread use of sibutramine in illicit
supplements (Phattanawasin et al., 2012) serves as one example of how
a withdrawn API could still make its way to the environment; hundreds
of nutritional supplements are known to have undeclared additives comprising known pharmaceuticals and unregistered analogs (Cohen, 2014;
USFDA, 2014).
Many APIs have limited occurrence data. Some of these are
enantiopure APIs, including those that are eutomers (the enantiomer
that possesses the desired pharmacologic effect); these stereoisomers
399
are constituents of their corresponding racemic drugs (which comprise
enantiomers in equal quantities). For example, the following APIs
are enantiopure eutomers of the racemic APIs listed in parentheses:
escitalopram (citalopram), esomeprazole (omeprazole), eszopiclone
(zopiclone), and levonorgestrel (norgestrel); the enantiomer lacking
the desired pharmacologic activity (distomer) might be ignored in environmental monitoring, even though it may be responsible for adverse
effects. Monitoring data may exist for the racemic API but not the
eutomer — or vice-versa; this may simply be a consequence of challenges
posed by chiral analysis of complex matrices. Other APIs are enantiomers
(but not necessarily eutomers), for example: dexmethylphenidate
(methylphenidate), levobupivacaine (bupivacaine), dilevalol (labetalol);
the racemic forms may have been targeted in environmental monitoring
but not their enantiomeric constituents. Enantiomers can add a level of
complexity in searching for published API occurrence data.
Lack of occurrence data for some APIs may be a consequence of
inadequate analytical methodologies, such as excessively high method
detection limits (e.g., 5-fluorouracil; valproic acid). Another factor that
can affect measured levels is the great natural variability associated
with sampling and variability in stream composition — especially sewage (Ort et al., 2010; Ort et al., 2014; Writer et al., 2013). This problem
is magnified by the variabilities associated with STP design, which impacts efficiency and which, in turn, is modulated by microbial activity
as affected by weather.
Other factors, which contribute to a dearth of occurrence data,
include the following: chemical instability or suspected short environmental half-life (e.g., carbidopa, chlordiazepoxide, cisplatin,
cyclobenzaprine, esmolol, isosorbide, nitroglycerin); natural variability
and error associated with sampling (Writer et al., 2013); or simply the
bona fide absence from the targeted matrix, such as via preferential
partitioning to solids (e.g., suspended particulates, sludge, sediments,
biofilms) thereby reducing their presence in a targeted dissolved phase
(e.g., clemastine, clindamycin, minocycline, paroxetine, tamoxifen).
Some matrices (such as sludge and sediments) are examined much
less frequently than aqueous samples. This may negatively bias the
occurrence data for those APIs that partition extensively to solids.
Some APIs are polypeptides and might therefore be expected to become
denatured (e.g., exenatide, goserelin, leuprolide, liraglutide, nafarelin,
octreotide, pramlintide) and therefore are purposefully omitted from
targeting, or they may pose analytical challenges (e.g., cyclobenzaprine),
especially in particular matrices (e.g., 5-fluororacil). Some APIs have
experienced dramatic declines in their usage, often because of widespread reported adverse reactions (e.g., thioridazine) or sometimes
because of widespread controversy (e.g., thiopental). Some are natural products or endogenous biochemicals whose monitoring may
not reflect exclusively the usage of medications (e.g., chloramphenicol, colchicine, dihydroquinidine, ergonovine, ergotamine, galantamine, levodopa, reserpine, scopolamine, vinblastine, vincristine).
The data for some pharmaceutical APIs may be convoluted with contributions from illegal drug use (e.g., chloramphenicol in aquaculture;
abusive use of flunitrazepam; sibutramine as an undeclared additive
to diet aids).
The occurrence and levels of an API can be highly influenced by its
primary or exclusive method of administration. For example, drugs
that are intended exclusively for topical administration (including
transdermal) can essentially be released to the environment in nearly
stoichiometric quantities during bathing (Daughton and Ruhoy, 2009).
Examples from BDDCS Class I include: betamethasone (and dexamethasone), bimatoprost, cortisone, hydrocortisone, imiquimod, lidocaine,
minoxidil, rotigotine, and triamcinolone. For these APIs, pharmacokinetics may not be a determining factor in their release to the environment. One ramification is that an API for exclusive topical usage
(depending on its rate of dermal absorption) might enter the environment in much greater quantities than an oral API — even one that is
highly excreted. APIs that experience high topical usage might make
high-probability targets for future monitoring.
400
C.G. Daughton / Science of the Total Environment 493 (2014) 392–404
Occurrence levels might also be elevated for some APIs even if they
undergo extensive metabolism. This can occur for medications that
have a higher propensity to accumulate unused (for example, those
with poor patient compliance or adherence, such as resulting from adverse reactions or use in treatment of conditions that do not display
overt symptoms) and later are disposed to sewers (Daughton, 2010a).
Disposal to sewers may be a factor that could account for excursions
or isolated reports of sporadically high levels in wastewaters.
Much published data on environmental occurrence results not
from formal environmental monitoring activities designed with sampling plans, but rather from research designed to verify new analytical
methodologies; the major objective of these studies is to demonstrate
the utility of new analytical methods — generally by acquiring analytical
figures of merit using isolated, unrepresentative samples (e.g., grab
samples) collected from various environmental matrices. Monitoring
data from multiple studies for a given API (or multiple APIs) usually cannot be directly compared because of the countless variables that impact
sampling and analysis, including the use of disparate and non-standard
methodologies, and even the effort that may or may not have been devoted to verifying molecular identity (e.g., via acquisition of accurate
mass and the use of certified standards). No attempt was made in this
assessment to distinguish a study or provide weighting to a study
according to any number of possible criteria, including the number of
samples collected, the sampling or analytical methodology or quality assurance (especially including the verification of analyte identification),
or geographic location, which may play a major role in dictating the
types and quantities of drugs consumed (e.g., as a result of contributions
from medical facilities, CAFOs, or manufacturing, or as a result of
prescribing customs, or season of year, which influences the usage of
certain medications); some drugs are used almost exclusively at hospitals (e.g., ifosfamide, methohexital) or predominantly at CAFOs and
therefore have limited geographic reach.
Finally, occurrence data can vary dramatically as a function of numerous factors associated with geography and governing boundaries,
including gross differences in seasons (e.g., solar irradiance and temperature, which modify both biological and physical processes that act
upon APIs, and seasonal distribution and incidence of diseases, which
affects the types and doses of APIs prescribed). Many medications are
not approved for use in all countries, and particular drugs may be withdrawn from markets in some countries but not others. Certain APIs may
be readily available OTC in some countries but only via prescription
in others. Likewise, the types and relative quantities of APIs can vary
dramatically across geographic locales as a function of prescribing
preferences, customs, and fads, as well as consumer preferences, beliefs,
and behaviors (such as compliance and adherence, or drug popularity as
influenced by consumer advertising). Recommended daily dose (which
largely reflects potency and bioavailability) can vary among countries
and often serves only as a guide to physicians. The age structures of
populations also dictate the distribution and quantities of the types of
APIs prescribed (with the incidence of polypharmacy increasing with
age). All of these factors can introduce large geographic discrepancies
in relative usage patterns and amounts, and thereby muddy the
comparison of occurrence data across locales, regions, or countries —
especially when the occurrence data are generated by disparate studies
using different sampling and analytical methodologies.
3. Results and conclusions
The environmental occurrence data for the APIs in the two selected
groups (BDDCS Class I and Class IV) as presented by Benet et al.
(2011) are compiled in Supplemental Tables S-1 and S-2, respectively.
These data are further organized into three somewhat subjective
groups (as described in Approach) according to whether the availability
of positive or negative occurrence data in the literature is Abundant,
Limited, or scarce (Paucity of data). The APIs in each of these three
groups for which occurrence data exceeded a threshold level of 1 μg/L
(or 1 mg/kg) — as compiled in Supplemental Tables S-1 and S-2 — are
summarized in Table 1 (for BDDCS Class I APIs) and Table 2 (for
BDDCS Class IV APIs).
3.1. Occurrence data from Table 1 (BDDCS Class I)
The following summarizes the findings for each of the three groups
of APIs (a total of 322) with respect to positive occurrence data:
Abundant occurrence data: A total of 57 APIs (18%) were in this group
(Supplemental Table S-1). Of these APIs, there were 27 (47%) with
data pointing to a routine occurrence exceeding 1 μg/L. And of
these, only 8 (14%) had data pointing to a routine occurrence
exceeding 5 μg/L or 1 mg/kg.
Limited occurrence data: A total of 41 APIs (13%) were in this group
(Supplemental Table S-1). Of these APIs, there were 5 (12%) with
data pointing to a routine occurrence exceeding 1 μg/L; these same
five also had data pointing to a routine occurrence exceeding 5 μg/L.
No study was located that reported an API level that exceeded
32.2 μg/L (i.e., for escitalopram) or 1 mg/kg (i.e., for zopiclone).
Paucity of occurrence data: A total of 224 APIs (69%) were in this
group (Supplemental Table S-1). Of these APIs, there were 11 (5%)
with but a few data pointing to the possibility of occurrence exceeding 1 μg/L. And of these, 8 (4%) had data pointing to the possibility of
occurrence exceeding 5 μg/L or 1 mg/kg. No study was located that
reported an API level that exceeded 15.4 μg/L (i.e., for indapamide)
or 9.3 mg/kg (i.e., for valproic acid).
The number of APIs for which no data were available (not yet
targeted in any study) totaled 176 (79% of the 224); the 224 APIs in
the Paucity of data group could each be examined for whether it
might be a Matthew Effect Orphaned Chemical — MEOC (as explained
in Daughton, 2014). Note that for the 53 highly prescribed APIs that
were first reported as possible MEOCs (Daughton, 2014), only 18 are
also captured among these 224 APIs in BDDCS Class I: benazepril,
carbidopa, colchicine, cyclobenzaprine, doxazosin, formoterol, hydralazine,
hydroxychloroquine, isosorbide, nitroglycerin, olmesartan medoxomil,
ondansetron, oxybutynin, ropinirole, sumatriptan, tamsulosin, terazosin,
and valacyclovir. This means that 206 [224 minus 18] of the APIs captured
in the Paucity of data group represent potentially new MEOCs. At the
least, these APIs for which occurrence data do not exist could serve as
targets for new monitoring studies.
3.2. Occurrence data from Table 2 (BDDCS Class IV)
The following summarizes the findings for each of the three groups
of APIs (a total of 52) with respect to positive occurrence data:
Abundant occurrence data: A total of 13 APIs (25%) were in this group
(Supplemental Table S-2). Of these APIs, the data for 10 (19%) pointed to a routine occurrence exceeding 1 μg/L or 1 mg/kg. Of these,
only 5 (10%) had data pointing to occurrence exceeding 5 μg/L
or 5 mg/kg. Ten of these APIs were antibiotics, and many of these
preferentially partitioned to solids.
Limited occurrence data: A total of 8 APIs (15%) were in this group
(Supplemental Table S-2). Of these APIs, there were 4 (8%) with
data pointing to a routine occurrence exceeding 1 μg/L; of these
four, 3 had levels exceeding 5 μg/L or 5 mg/kg. No study was located
that reported an API level that exceeded 8.9 μg/L (i.e., for chlorothiazide) or 20.1 μg/kg (i.e., for chlorthalidone).
Paucity of occurrence data: A total of 31 APIs (60%) were in this group
(Supplemental Table S-2). None had data pointing to the possibility
of occurrence exceeding 1 μg/L.
The number of APIs for which no data were available (not yet
targeted in any study) totaled 22 (71% of 31); most of the remainder
had data from only one or two studies. The 31 APIs in the Paucity of
data group could each be examined for whether it might be a MEOC.
C.G. Daughton / Science of the Total Environment 493 (2014) 392–404
Note that for the 53 highly prescribed APIs that were first reported as
possible MEOCs (Daughton, 2014), only 3 are also captured among
these 31 APIs in BDDCS Class IV: cefdinir, phenazopyridine, and
nitrofurantoin (chlorthalidone is listed under Limited data in the study
here). This means that 28 of the APIs captured in the Paucity of data
group represent potential new MEOCs. At the least, these APIs for
which occurrence data do not exist could serve as targets for new
monitoring studies.
3.3. Overall comparison of occurrence data from BDDCS Class I and Class IV
For this study, BDDCS Class I and Class IV comprised very disparate
numbers of APIs: 322 for Class I and 52 for Class IV; this discrepancy
was discussed earlier in Section 2.1 (Proxy measure for API excretion:
the BDDCS). Even so, one obvious commonality between the two classes
is the large number of APIs for which no occurrence data were available
(i.e., those having not yet been targeted in any published study). These
numbers totaled 176 of 322 (55%) for Class I — and 22 of 52 (42%) for
Class IV. So there were no occurrence data available (as of 8 May
2014) for roughly half of all the APIs subject to this examination.
The number of APIs with data pointing to elevated levels were
41 (13%) of the total number in Class I — or 41 of the 108 total (38%)
having data. The number of APIs with data pointing to elevated levels
were 14 (27%) of the total number in Class IV — or 14 of the 21 total
(67%) having data. This weight of evidence points to a possible trend
of higher incidence of elevated levels among BDDCS Class IV APIs; a
disproportionate number of these data, however, derived from antibiotics that partition to solids. Of the APIs in the Abundant and limited
data groups having the highest levels in solids, six were Class IV
APIs (ciprofloxacin, erythromycin stearate, fleroxacin, norfloxacin,
roxithromycin, and valsartan, with maximum levels ranging from 1 to
5.6 mg/kg), while only two were Class I (minocycline and zopiclone,
with a maximum level of 1 mg/kg); these levels are roughly 3 orders
of magnitude higher than the highest levels reported for aqueous
samples (probably because of the surface-concentration effected by
sorption and because levels in solids are often reported on a dryweight basis). Higher occurrence levels for Class IV drugs would be
expected not just by their poor metabolism but also by the need to
administer higher doses (less potency), which leads to greater direct
excretion (poor absorption).
The elevated levels among certain Class I APIs could be caused by any
number of reasons, including the following: exceptionally high usage
rates (e.g., several of these APIs are among the more highly prescribed
drugs: diclofenac, diltiazem, metoprolol, propranolol); direct disposal
to sewers (including consumers and hospitals, where unit-dose packaged injectables are frequently sent to sewers — a common practice,
for example, with hydromorphone); substantial contributions from
hospitals/healthcare facilities (e.g., ifosfamide); CAFOs (many antibiotics);
possible illegal agricultural usage (e.g., chloramphenicol); abuse
or recreational use (e.g., flunitrazepam, hydroxyzine, methadone,
methylphenidate, oxycodone); exceptional environmental half-lives
(e.g., clofibric acid); bias from time of day or season of sample collection
(e.g., oseltamivir); geographic distribution of disease (e.g., zidovudine);
and pharmacokinetics characterized by extensive metabolism but coupled
with extensive excretion of reversible conjugates (e.g., zidovudine).
The data can also be examined from the other end of the spectrum —
APIs with data of absence (negative data). Among the Class I APIs
that have been targeted by monitoring, it is readily evident that
at least 27 (8%) only have data reflecting very low levels (ng/L) or
were not detectable: alprenolol, ambroxol, betaxolol, bromocriptine,
cilazapril, clemastine, clomipramine, dexamethasone, dextromethorphan, doxazosin, duloxetine, fentanyl, finasteride, fluorouracil,
fluvoxamine, ifosfamide, irinotecan, maprotiline, methylphenidate,
midazolam, norgestimate, prochlorperazine, ribavirin, triamcinolone acetonide, triamcinolone, and vinorelbine. Of course, trends
establishing data of absence can only be strengthened with additional
401
targeted monitoring data; such data can be made more compelling
but never be claimed as certain.
Likewise, among the Class IV APIs that have been targeted by monitoring, only about 5 (10%) have data supporting very low levels (ng/L)
or were not detectable: cefdinir, iopanoic acid, medroxyprogesterone
acetate, megestrol acetate, and meropenem. Furthermore, many of the
Class I APIs in the Abundance group are frequently reported with
mixed or conflicting findings (e.g., low levels or only sporadically at appreciable levels; bromazepam and secobarbital are but two examples).
In contrast, all of the Class IV APIs in the Abundance group were frequently and consistently reported at appreciable or substantial levels.
Since Class IV APIs may have the higher probability of elevated occurrence levels, they might serve as the more likely targets for future
monitoring — especially those that are possible MEOCs.
APIs with compelling data of absence have significant implications
with respect to medical prescribing. The loadings of these APIs in the
environment would possibly be influenced the least as a result of ultimate use by patients. This points to the importance of diligence in the
reporting of negative occurrence data for APIs from environmental
monitoring (Daughton, 2014). APIs with abundant data of absence
have the potential for the lowest environmental footprints (assuming
direct disposal to sewers is avoided and bioactive metabolites are not
a concern).
The weight of evidence (including the absence of evidence) that was
revealed in this examination of environmental occurrence data tends to
support the possible utility of using the BDDCS as a means of quickly
informing medical practitioners as to the potential for environmental
impact of an API. A more in-depth study would be needed to strengthen
the trends that seemed to emerge — namely BDDCS Class I APIs being
associated with reduced environmental presence compared with Class
IV APIs. Additional APIs would need to be evaluated from both classes
with respect to environmental occurrence. Yet more and ongoing literature searching is required for the substantial numbers of APIs that lack
data. Alternatively, decisions should be made with respect to the possible MEOCs as to whether their targeted monitoring might be warranted.
Additionally, consideration should be given to an analogous examination of BDDCS Class II APIs and Class III APIs (which are also poorly
metabolized, like Class IV) to see if there are correlations regarding
their environmental occurrence.
4. Future directions
An improved ability to predict the types and quantities of APIs that
have the potential to enter the environment would certainly help
guide the targeting of APIs to monitor in the environment. Access to
real-time, geographic usage data is the major limitation to quantifying
the scope (types, amounts, and locations) of API sources (Daughton,
2013). Comprehensive commercial informatics services are available
in some countries. These databases compile detailed data on prescription sales, dispensing, and demographics, but access is often fee-based,
which usually precludes their utility for modeling and predictive purposes. Even then, it is unknown what portions of dispensed drugs are
ultimately used versus those that may be indefinitely stockpiled or
disposed by end-users. Furthermore, the temporal delay between
times of dispensing and ultimate use can range into the years. Disparities in spatial disconnects between the location of prescription sale
and the geographic locale where the drug is ultimately used (due to
population mobility) further complicates modeling; unknown portions
of certain drugs are ultimately used in regions or countries where they
were not originally dispensed, and a certain portion of some drugs
that are legally dispensed only by prescription are widely purchased
illegally. These and many other problems that impinge on the utility
of modeling for predicting levels of APIs in the environment have
been discussed (Daughton, 2013). Empirical monitoring data are critical
for revealing which APIs to target for pollution prevention efforts and
402
C.G. Daughton / Science of the Total Environment 493 (2014) 392–404
for verifying the effectiveness of any prevention, control, or mitigation
measures that have been implemented.
Pollution prevention approaches for reducing the entry of APIs to the
environment must accommodate the interconnected whole — with the
environment and patients essentially being treated as a single, integral
system. Measures that might be protective for one may pose risks
for the other. These tradeoffs require balancing — while at the same
time ensuring that any alterations to the administration or practice
of healthcare do not jeopardize human health or reduce economic
efficiency. Ensuring an evidence-based approach for drug and dose
selection is critical. An integrated approach will eventually require
collaboration between environmental scientists and healthcare
professionals — two groups that have historically never communicated;
it also will require cooperation among disparate federal and state agencies involved with protection of the environment, administration of
medical care, and regulation of medication sales and disposal (especially
controlled substances).
EDSP marks the first time that API pharmacokinetics (using the
BDDCS as a ready proxy) has been examined as a factor that could be
used to guide decisions involving prescribing, dispensing, and end-use
of drugs for the purpose of minimizing environmental impact. Changing
the prescribing behavior of physicians would certainly be a major
challenge. EDSP would represent the very first attempt at providing
prescribers, dispensers, and users (patients) with pollution prevention
information to consider in their selection of medications or dosages.
The proposed approach would represent the first of undoubtedly multiple future steps required for changing behavior. At the least, the EDSP
concept would serve to raise awareness that while excretion may represent the major source of most APIs in the environment, these levels can
nonetheless be actively reduced — with no added infrastructure costs
(such as entailed with improved wastewater treatment). The EDSP
could mesh well with the emerging clinical movement of “conservative
prescribing” (Schiff et al., 2011).
The proposed EDSP approach could be made even more effective
with the eventual widespread implementation of personalized medicine (“precision” medicine), which is being accelerated with advances
in pharmacogenomics. Not until the last couple of years had consideration been given to tailoring medications to patients with the intention
of lowering the excretion of parent APIs or bioactive metabolites
(see: Daughton and Ruhoy, 2011). By appropriate evaluation of the PK
characteristics (absorption, distribution, metabolism, and excretion)
for a particular API, better-informed decisions could be made regarding
those APIs in a specific therapeutic class having less potential for
environmental impact via excretion. For example, prescribing certain
APIs could be avoided or reduced for individuals having non-optimal
metabolism (e.g., heightened excretion of the API), for those taking
other medications that inhibit the absorption or metabolism of the
API, or for those who are simply poor therapeutic responders.
Excretion profiles would be useful not just for guiding the selection
of drugs for prescribing but could also prove very useful for guiding
decisions regarding whether a particular API could be prudently disposed to sewers. Excretion profiles could be used to assess the potential
for whether the disposal of a particular drug would contribute significantly to the API's overall environmental loading. For example, some
APIs are extensively excreted unchanged. For these APIs, disposal to
sewers might add only small incremental portions to the already comparatively high ambient environmental levels continually contributed
by excretion. In contrast, for those APIs that are extensively metabolized
(little API is excreted unchanged or as reversible conjugates), sewer
disposal holds the potential for contributing significant portions to ambient levels. This aspect of drug disposal has been under-recognized,
especially in the formulation of regulations and guidance aimed at curbing sewer disposal. Despite the growing focus in the US on end-of-pipe
pollution control programs for collecting leftover, unwanted drugs (and
shunning their disposal to sewers), sewer disposal may well be the best
option for the disposal of certain drugs (i.e., those excreted unchanged).
This is especially true for those APIs with high acute toxicity (i.e., those
with single-dose lethality) and those subject to diversion and abuse
(certain synthetic opiates are but one example). Failure to immediately
dispose or secure these leftover drugs (and associated delivery devices,
such as used transdermal patches) is a documented cause of deaths in
infants and young children (Daughton, 2010a; Daughton and Ruhoy,
2009); this has been a concern of the FDA regarding guidance for the
sewer disposal of certain medications (USFDA, 2009). For those highly
hazardous drugs that are extensively excreted unchanged, disposal of
leftovers to sewers might continue to be the best means of preventing
fatal poisonings. Furthermore, if sewer-disposal of a highly hazardous
drug contributes only a small portion of the overall environmental
burden of its API, then disposal may prove to have only nominal
added impact on the aquatic environment.
The EDSP concept would need to be translated into clinical practice.
Currently, the medical community receives little exposure to information regarding the environmental impact of their professions; environmental impact is not routinely incorporated in medical training. In the
US, expertise in outreach medical education for translation, dissemination, and implementation resides at the AHRQ Effective Health Care
Program (http://www.effectivehealthcare.ahrq.gov/), which operates
under the HHS Agency for Healthcare Research and Quality (AHRQ).
The AHRQ is the “lead Federal agency charged with improving the quality, safety, efficiency, and effectiveness of health care for all Americans”,
primarily through evidence-based decision making. The AHRQ has a
number of mechanisms for communicating with doctors: spanning
the spectrum from on-line continuing education to one-on-one inoffice outreach visits with physicians (via the National Resource Center
for Academic Detailing, NaRCAD: http://www.narcad.org/). One of few
examples of pollution prevention being considered for reducing drug
waste was the recognition by the pharmacy community for the need
to develop actions to reduce the incidence of leftover drugs rather
than focus on waste disposal, as formally proposed in 2009 by the
National Association of Boards of Pharmacy: “Recommendation 3:
Work with Appropriate Entities to Research Methods that Reduce the
Amount of Unused Medications” (NABP, 2009).
A major objective of the work reported here is to foster increased
recognition of the potential role for pollution prevention rather than
pollution mitigation — particularly for reducing the many actions and
behaviors in the healthcare communities that lead to the unnecessary
and imprudent use of medications and generation and accumulation
of avoidable drug waste. Prevention continues to remain an unused approach for dealing with the dual problems of drug waste and excreted
residues and their resulting impacts on both human and environmental
health.
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.scitotenv.2014.06.013.
Acknowledgments
U.S. EPA Notice: The views expressed in this article are those of the
author and do not necessarily reflect the views or policies of the U.S.
Environmental Protection Agency. Mention of trade names or commercial products does not constitute endorsement or recommendation
for use. This work was supported by EPA-ORD's Pathfinder Innovation
Program, which was launched in October, 2010. Greatly appreciated
were the assistance of MST Scuderi (SEEP, U.S. EPA, Las Vegas) in the
maintenance of the EndNote bibliographic database and the technical
review of this manuscript by Dr. Tamara Sorell (National Risk Practice
Lead, Brown and Caldwell, Andover, MA).
References
Barnes KK, Kolpin DW, Furlong ET, Zaugg SD, Meyer MT, Barber LB. A national reconnaissance of pharmaceuticals and other organic wastewater contaminants in the United
States — I) groundwater. Sci Total Environ 2008;402:192–200. http://dx.doi.org/
10.1016/j.scitotenv.2008.04.028.
C.G. Daughton / Science of the Total Environment 493 (2014) 392–404
Baum R, Luh J, Bartram J. Sanitation: a global estimate of sewerage connections without
treatment and the resulting impact on MDG progress. Environ Sci Technol 2013;47:
1994–2000. http://dx.doi.org/10.1021/es304284f.
Benet LZ. The role of BCS (biopharmaceutics classification system) and BDDCS
(biopharmaceutics drug disposition classification system) in drug development.
J Pharm Sci 2013;102:34–42. http://dx.doi.org/10.1002/jps.23359.
Benet LZ, Broccatelli F, Oprea TI. BDDCS applied to over 900 drugs. AAPS J 2011;13:
519–47. http://dx.doi.org/10.1208/s12248-011-9290-9.
Blair BD, Crago JP, Hedman CJ, Treguer RJF, Magruder C, Royer LS, et al. Evaluation of a
model for the removal of pharmaceuticals, personal care products, and hormones
from wastewater. Sci Total Environ 2013;444:515–21. http://dx.doi.org/10.1016/
j.scitotenv.2012.11.103.
Bond GR, Woodward RW, Ho M. The growing impact of pediatric pharmaceutical poisoning. J Pediatr 2012;160:265-270.e1. http://dx.doi.org/10.1016/j.jpeds.2011.07.042.
Budnitz DS, Salis S. Preventing medication overdoses in young children: an opportunity
for harm elimination. Pediatrics 2011;127:e1597–9. http://dx.doi.org/10.1542/
peds.2011-0926.
Burghardt LC, Ayers JW, Brownstein JS, Bronstein AC, Ewald MB, Bourgeois FT. Adult prescription drug use and pediatric medication exposures and poisonings. Pediatrics
2013;132:1–10. http://dx.doi.org/10.1542/peds.2012-2978.
Charkoftaki G, Dokoumetzidis A, Valsami G, Macheras P. Biopharmaceutical classification
based on solubility and dissolution: a reappraisal of criteria for hypothesis models in
the light of the experimental observations. Basic Clin Pharmacol Toxicol 2010;106:
168–72. http://dx.doi.org/10.1111/j.1742-7843.2009.00506.x.
Coday BD, Yaffe BGM, Xu P, Cath TY. Rejection of trace organic compounds by forward
osmosis membranes: a literature review. Environ Sci Technol 2014;48:3612–24.
http://dx.doi.org/10.1021/es4038676.
Cohen PA. Hazards of hindsight — monitoring the safety of nutritional supplements.
N Engl J Med 2014;370:1277–80. http://dx.doi.org/10.1056/NEJMp1315559.
Curfman GD, Morrissey S, Drazen JM. High-value health care — a sustainable proposition.
N Engl J Med 2013;369:1163–4. http://dx.doi.org/10.1056/NEJMe1310884.
Custodio JM, Wu CY, Benet LZ. Predicting drug disposition, absorption/elimination/
transporter interplay and the role of food on drug absorption. Adv Drug Deliv
Rev 2008;60:717–33. http://dx.doi.org/10.1016/j.addr.2007.08.043.
Daneshvar A. Source, occurrence, and fate of pharmaceuticals in natural waters. Doctoral
dissertation Uppsala, Sweden: Swedish University of Agricultural Sciences; 2012.
p. 74 [Available on: http://pub.epsilon.slu.se/9100/ (accessed 8 April 2014)].
D'Ascenzo G, Di Corcia A, Gentili A, Mancini R, Mastropasqua R, Nazzari M, et al. Fate of
natural estrogen conjugates in municipal sewage transport and treatment facilities.
Sci Total Environ 2003;302:199–209. http://dx.doi.org/10.1016/S0048-9697(02)
00342-X.
Daughton CG. Pharmaceuticals and personal care products in the environment: overarching
issues and overview. In: Daughton CG, Jones-Lepp TL, editors. Pharmaceuticals and
personal care products in the environment: scientific and regulatory issues. Washington,
DC: American Chemical Society; 2001. p. 2–38. [chapter 1].
Daughton CG. Environmental stewardship and drugs as pollutants. Lancet 2002;360:
1035–6. http://dx.doi.org/10.1016/S0140-6736(02)11176-7.
Daughton CG. Cradle-to-cradle stewardship of drugs for minimizing their environmental
disposition while promoting human health. I. Rationale for and avenues toward
a green pharmacy. Environ Health Perspect 2003;111:757–74. http://dx.doi.org/
10.1289/ehp.5947.
Daughton CG. Pharmaceuticals as environmental pollutants: the ramifications for human
exposure. In: Heggenhougen K, Quah S, editors. International encyclopedia of public
health. Oxford, England: Academic Press; 2008. p. 66–102.
Daughton CG. Pharmaceutical ingredients in drinking water: overview of occurrence and
significance of human exposure. In: Halden RU, editor. Washington, DC: American
Chemical Society; 2010a. [chapter 2].
Daughton CG. Drugs and the environment: stewardship & sustainability. NERL-LV-ESD
10/081, EPA/600/R-10/106. Las Vegas, NV: National Exposure Research Laboratory,
Environmental Sciences Division, US EPA; 2010b. p. 196 [Available on: http://www.
epa.gov/nerlesd1/bios/daughton/APM200-2010.pdf (accessed 8 April 2014)].
Daughton CG. Pharmaceuticals in the environment: sources and their management. In:
Petrovic M, Perez S, Barcelo D, editors. Analysis, removal, effects and risk of pharmaceuticals in the water cycle — occurrence and transformation in the environment.
Elsevier; 2013. p. 37–69. [chapter 2].
Daughton CG. The Matthew Effect and widely prescribed pharmaceuticals lacking environmental monitoring: case study of an exposure-assessment vulnerability. Sci
Total Environ 2014;466–467:315–25.
Daughton CG, Ruhoy IS. Environmental footprint of pharmaceuticals: the significance
of factors beyond direct excretion to sewers. Environ Toxicol Chem 2009;28:
2495–521. http://dx.doi.org/10.1897/08-382.1.
Daughton CG, Ruhoy IS. Green pharmacy and PharmEcovigilance: prescribing and the planet.
Expert Rev Clin Pharmacol 2011;4:211–32. http://dx.doi.org/10.1586/ecp.11.6.
Daughton CG, Ruhoy IS. Lower-dose prescribing: minimizing “side effects” of pharmaceuticals on society and the environment. Sci Total Environ 2013;443:324–37. http://
dx.doi.org/10.1016/j.scitotenv.2012.10.092.
Daughton CG, Scuderi MST. Pharmaceuticals and personal care products (PPCPs):
relevant literature. Las Vegas, NV: U.S. Environmental Protection Agency; 2014
[a comprehensive bibliographic database of literature references; first implemented
19 February 2008].
Deblonde T, Hartemann P. Environmental impact of medical prescriptions: assessing the
risks and hazards of persistence, bioaccumulation and toxicity of pharmaceuticals.
Public Health 2013;127:312–7. http://dx.doi.org/10.1016/j.puhe.2013.01.026.
Debroux J-F, Soller JA, Plumlee MH, Kennedy LJ. Human health risk assessment of nonregulated xenobiotics in recycled water: a review. Hum Ecol Risk Assess 2012;18:
517–46. http://dx.doi.org/10.1080/10807039.2012.672883.
403
Deo RP. Pharmaceuticals in the surface water of the USA: a review. Curr Environ Health
Rep 2014;1:113–22. http://dx.doi.org/10.1007/s40572-014-0015-y.
Deo RP, Halden RU. Pharmaceuticals in the built and natural water environment of the
United States. Water 2013;5:1346–65. http://dx.doi.org/10.3390/w5031346.
Desbrow C, Routledge EJ, Brighty GC, Sumpter JP, Waldock M. Identification of estrogenic
chemicals in STW effluent. 1. Chemical fractionation and in vitro biological screening.
Environ Sci Technol 1998;32:1549–58.
Focazio MJ, Kolpin DW, Barnes KK, Furlong ET, Meyer MT, Zaugg SD, et al. A national
reconnaissance for pharmaceuticals and other organic wastewater contaminants in
the United States — II) untreated drinking water sources. Sci Total Environ 2008;
402:201–16. http://dx.doi.org/10.1016/j.scitotenv.2008.02.021.
Gao L, Shi Y, Li W, Niu H, Liu J, Cai Y. Occurrence of antibiotics in eight sewage treatment
plants in Beijing, China. Chemosphere 2012;86:665–71. http://dx.doi.org/10.1016/
j.chemosphere.2011.11.019.
García-Galán MJ, González Blanco S, López Roldán R, Díaz-Cruz S, Barceló D. Ecotoxicity
evaluation and removal of sulfonamides and their acetylated metabolites during
conventional wastewater treatment. Sci Total Environ 2012;437:403–12. http://
dx.doi.org/10.1016/j.scitotenv.2012.08.038.
Glassmeyer ST, Hinchey EK, Boehme SE, Daughton CG, Ruhoy IS, Conerly O, et al. Disposal
practices for unwanted residential medications in the United States. Environ Int
2009;35:566–72. http://dx.doi.org/10.1016/j.envint.2008.10.007.
Hardikar SR, Bhosale AV, Chonkar AD, Saindane RA. Dissolution and bioavailability
enhancement of carbamazepine. Int J Pharm Pharm Sci 2013;5:395–400.
Hermening A, Grafe A-K, Baktir G, Mutschler E, Spahn-Langguth H. Gemfibrozil and its
oxidative metabolites: quantification of aglycones, acyl glucuronides, and covalent
adducts in samples from preclinical and clinical kinetic studies. J Chromatogr B
2000;741:129–44. http://dx.doi.org/10.1016/S0378-4347(00)00041-4.
Hughes SR, Kay P, Brown LE. Global synthesis and critical evaluation of pharmaceutical
data sets collected from river systems. Environ Sci Technol 2013;47:661–77. http://
dx.doi.org/10.1021/es3030148.
Jamei M, Turner D, Yang J, Neuhoff S, Polak S, Rostami-Hodjegan A, et al. Populationbased mechanistic prediction of oral drug absorption. AAPS J 2009;11:225–37.
http://dx.doi.org/10.1208/s12248-009-9099-y.
Jones OAH, Voulvoulis N, Lester JN. Human pharmaceuticals in wastewater treatment
processes. Crit Rev Environ Sci Technol 2005;35:401–27. http://dx.doi.org/10.1080/
10643380590956966.
Karanje R, Bhavsar Y, Jahagirdar K, Bhise K. Formulation and development of extendedrelease micro particulate drug delivery system of solubilized rifaximin. AAPS
PharmSciTech 2013;14:639–48. http://dx.doi.org/10.1208/s12249-013-9949-x.
Kolpin DW, Furlong ET, Meyer MT, Thurman EM, Zaugg SD, Barber LB, et al. Pharmaceuticals,
hormones, and other organic wastewater contaminants in U.S. streams, 1999–2000:
a national reconnaissance. Environ Sci Technol 2002;36:1202–11. http://dx.doi.org/
10.1021/es011055j.
Kumar V, Johnson AC, Nakada N, Yamashita N, Tanaka H. De-conjugation behavior of
conjugated estrogens in the raw sewage, activated sludge and river water. J Hazard
Mater 2012;227–228:49–54. http://dx.doi.org/10.1016/j.jhazmat.2012.04.078.
Lienert J, Burki T, Escher BI. Reducing micropollutants with source control: substance
flow analysis of 212 pharmaceuticals in faeces and urine. Water Sci Technol 2007;
56:87–96.
Liu Z-h, Kanjo Y. An innovative analytical method for estrogen sulfates without deconjugation procedure. KSCE J Civ Eng 2012;16:919–24. http://dx.doi.org/10.1007/
s12205-012-1666-6.
Luo Y, Guo W, Ngo HH, Nghiem LD, Hai FI, Zhang J, et al. A review on the occurrence of
micropollutants in the aquatic environment and their fate and removal during wastewater treatment. Sci Total Environ 2014;473–474:619–41. http://dx.doi.org/10.1016/
j.scitotenv.2013.12.065.
Macheras P, Karalis V, Valsami G. Keeping a critical eye on the science and the regulation
of oral drug absorption: a review. J Pharm Sci 2013;102:3018–36. http://dx.doi.org/
10.1002/jps.23534.
Mackay D, Hughes DM, Romano ML, Bonnell M. The role of persistence in chemical
evaluations. Integr Environ Assess Manag 2014. http://dx.doi.org/10.1002/ieam.1545.
[in press].
McConnell EL, Fadda HM, Basit AW. Gut instincts: explorations in intestinal physiology
and drug delivery. Int J Pharm 2008;364:213–26. http://dx.doi.org/10.1016/
j.ijpharm.2008.05.012.
NABP2008–2009 report of the task force on medication collection programs. Mount
Prospect, IL: National Association of Boards of Pharmacy; 2009. p. 7. [Available on:
https://www.nabp.net/news/2008-2009-report-of-the-task-force-on-medicationcollection-programs; http://www.nabp.net/news/assets/08TF_Med_Collection_
Programs.pdf (accessed 8 April 2014].
ONDCP. National drug control strategy. Rockville, MD: Office of the National Drug Control
Policy; 2013. p. 95 (Available on: http://www.whitehouse.gov/ondcp/national-drugcontrol-strategy; http://www.whitehouse.gov//sites/default/files/ondcp/policy-andresearch/ndcs_2013.pdf (accessed 8 April 2014)).
Ort C, Lawrence MG, Rieckermann J, Joss A. Sampling for pharmaceuticals and personal
care products (PPCPs) and illicit drugs in wastewater systems: are your conclusions
valid? A critical review. Environ Sci Technol 2010;44:6024–35. http://dx.doi.org/
10.1021/es100779n.
Ort C, Eppler JM, Scheidegger A, Rieckermann J, Kinzig M, Sörgel F. Challenges of surveying wastewater drug loads of small populations and generalisable aspects on optimizing monitoring design. Addiction 2014;109:472–81. http://dx.doi.org/10.1111/
add.12405.
Panter GH, Thompson RS, Beresford N, Sumpter JP. Transformation of a non-oestrogenic
steroid metabolite to an oestrogenically active substance by minimal bacterial
activity. Chemosphere 1999;38:3579–96. http://dx.doi.org/10.1016/S00456535(98)00572-4.
404
C.G. Daughton / Science of the Total Environment 493 (2014) 392–404
Pham-The H, Garrigues T, Bermejo M, González-Álvarez I, Monteagudo MC, CabreraPérez MÁ. Provisional classification and in silico study of biopharmaceutical system
based on Caco-2 cell permeability and dose number. Mol Pharm 2013;10:2445–61.
http://dx.doi.org/10.1021/mp4000585.
Phattanawasin P, Sotanaphun U, Sukwattanasinit T, Akkarawaranthorn J, Kitchaiya S.
Quantitative determination of sibutramine in adulterated herbal slimming formulations by TLC-image analysis method. Forensic Sci Int 2012;219:96–100. http://
dx.doi.org/10.1016/j.forsciint.2011.12.004.
Rice J, Westerhoff P, Wutich A. Assessment of de facto wastewater reuse across the USA:
trends between 1980 and 2008. Environ Sci Technol 2013;47:11099–105. http://
dx.doi.org/10.1021/es402792s.
Ruhoy IS, Daughton CG. Beyond the medicine cabinet: an analysis of where and why
medications accumulate. Environ Int 2008;34:1157–69. http://dx.doi.org/10.1016/
j.envint.2008.05.002.
Schiff GD, Galanter WL, Duhig J, Lodolce AE, Koronkowski MJ, Lambert BL. Principles
of conservative prescribing. Arch Intern Med 2011;171:1433–40. http://dx.doi.org/
10.1001/archinternmed.2011.256.
Schiller C, Fröhlich CP, Giessmann T, Siegmund W, Mönnikes H, Hosten N, et al. Intestinal
fluid volumes and transit of dosage forms as assessed by magnetic resonance imaging. Aliment Pharmacol Ther 2005;22:971–9. http://dx.doi.org/10.1111/j.13652036.2005.02683.x.
Snyder SA, Adham S, Redding AM, Cannon FS, DeCarolis J, Oppenheimer J, et al. Role of
membranes and activated carbon in the removal of endocrine disruptors and pharmaceuticals. Desalination 2007;202:156–81. http://dx.doi.org/10.1016/j.desal.2005.12.052.
Stockholm County Council. Environmentally classified pharmaceuticals. Stockholm,
Sweden: Stockholm City Council; 2012. p. 48 (Available on: http://www.janusinfo.se/
Global/Miljo_och_lakemedel/miljobroschyr_2011_uppslag_eng.pdf (accessed 8 April
2014)).
Sui Q, Huang J, Deng S, Chen W, Yu G. Seasonal variation in the occurrence and removal of
pharmaceuticals and personal care products in different biological wastewater treatment processes. Environ Sci Technol 2011;45:3341–8. http://dx.doi.org/10.1021/
es200248d.
Ternes TA, Kreckel P, Mueller J. Behaviour and occurrence of estrogens in municipal sewage treatment plants — II. Aerobic batch experiments with activated sludge. Sci Total
Environ 1999;225:91–9. http://dx.doi.org/10.1016/S0048-9697(98)00335-0.
USFDA. Disposal of unused medicines: what you should know. list of medicines recommended for disposal by flushing [revised November 2013] Rockville, MD: US
Food and Drug Administration; 2009. Available on: http://www.fda.gov/Drugs/
ResourcesForYou/Consumers/BuyingUsingMedicineSafely/EnsuringSafeUseofMedicine/
SafeDisposalofMedicines/ucm186187.htm (accessed 8 April 2014).
USFDA. FDA's medication health fraud page: list of tainted dietary supplements. Silver Spring, MD: US Food and Drug Administration, Center for Drug Evaluation
and Research; 2014 [Available on: http://www.accessdata.fda.gov/scripts/sda/
sdNavigation.cfm?filter=&sortColumn=1d&sd=tainted_supplements_cder&page=1;
http://www.fda.gov/Drugs/ResourcesForYou/Consumers/BuyingUsingMedicineSafely/
MedicationHealthFraud/default.htm (accessed 8 April 2014)].
van der Aa NGFM, Dijkman E, Bijlsma L, Emke E, van de Ven BM, Van Nuijs ALN, et al.
Drugs of abuse and tranquilizers in Dutch surface waters, drinking water and
wastewater: results of screening monitoring 2009. Bilthoven, The Netherlands:
National Institute for Public Health and the Environment (RIVM); 2011. p. 90
[703719064/2010, Available on: http://dare.uva.nl/record/391403 (accessed 8
April 2014)].
Verlicchi P, Al Aukidy M, Zambello E. Occurrence of pharmaceutical compounds in
urban wastewater: removal, mass load and environmental risk after a secondary
treatment — a review. Sci Total Environ 2012;429:123–55. http://dx.doi.org/10.1016/
j.scitotenv.2012.04.028.
Wennmalm Ǻ, Gunnarsson B. Experiences with the Swedish environment classification
scheme. In: Kümmerer K, Hempel M, editors. Green and sustainable pharmacy. Berlin
Heidelberg, Germany: Springer-Verlag; 2010. p. 243–9. [chapter 16].
White RE, Evans DC, Hop CECA, Moore DJ, Prakash C, Surapaneni S, et al. Radiolabeled
mass-balance excretion and metabolism studies in laboratory animals: a commentary
on why they are still necessary. Xenobiotica 2013;43:219–25. http://dx.doi.org/10.
3109/00498254.2012.706724.
Williams RT, Cook JC. Exposure to pharmaceuticals present in the environment. Drug Inf J
2007;41:133–41.
Writer JH, Antweiler RC, Ferrer I, Ryan JN, Thurman EM. In-stream attenuation of neuroactive pharmaceuticals and their metabolites. Environ Sci Technol 2013;47:9781–90.
http://dx.doi.org/10.1021/es402158t.
Wu C-Y, Benet LZ. Predicting drug disposition via application of BCS: transport/absorption/elimination interplay and development of a biopharmaceutics drug disposition
classification system. Pharm Res 2005;22:11–23. http://dx.doi.org/10.1007/s11095004-9004-4.
Yan Q, Gao X, Chen Y-P, Peng X-Y, Zhang Y-X, Gan X-M, et al. Occurrence, fate and ecotoxicological assessment of pharmaceutically active compounds in wastewater and
sludge from wastewater treatment plants in Chongqing, the Three Gorges Reservoir
Area. Sci Total Environ 2014;470–471:618–30. http://dx.doi.org/10.1016/j.scitotenv.
2013.09.032.
Zhang H, Liu P, Feng Y, Yang F. Fate of antibiotics during wastewater treatment and
antibiotic distribution in the effluent-receiving waters of the Yellow Sea, northern China. Mar Pollut Bull 2013;73:282–90. http://dx.doi.org/10.1016/j.marpolbul.
2013.05.007.
Zhou JL, Zhang ZL, Banks E, Grover D, Jiang JQ. Pharmaceutical residues in wastewater
treatment works effluents and their impact on receiving river water. J Hazard
Mater 2009;166:655–61. http://dx.doi.org/10.1016/j.jhazmat.2008.11.070.